Benjamin Hansen, Daniel I. Rees, Joseph J. Sabia

AJHE NewsBrief: Have Cigarette Taxes Lost their Bite? New Estimates of the Relationship between Cigarette Taxes and Youth Smoking

Benjamin Hansen, University of Oregon, IZA and NBER
Daniel I. Rees, University of Colorado Denver and IZA
Joseph J. Sabia, University of New Hampshire and IZA

American Journal of Health Economics 3(1): 60–75, 2017

Since the late 1990s, the youth smoking participation rate has fallen dramatically, and is now at record-low levels.  According to Youth Risky Behavior Survey (YRBS) data, 36 percent of high school students reported smoking in the previous month in 1997.  By 2015, only 11 percent of high school students reported smoking in the previous month.

Over the same period, cigarette taxes increased substantially.  In 1997, the combined state and federal tax was, on average, $0.62 per pack.  By 2014, this figure had gone up to $2.56, and taxes made up almost 44% of the total price of a pack of cigarettes1.

The two concurrent trends described above raise a natural question: Were cigarette tax hikes responsible for the decline in youth smoking?  Many experts and policymakers think that they were, but there were other factors, both demographic and policy-related, that also could have also influenced the decision to take up (or not to take up) smoking.  In our article, “Have Cigarette Taxes Lost Their Bite?  New Estimates on the Relationship Between Cigarette Taxes and Youth Smoking”, recently published by the American Journal of Health Economics, we exploited differences in the timing and magnitude of state-level cigarette tax increases from 1993-2013 to assess their impact on youth smoking.

The academic debate on cigarette taxes and youth smoking has been going on for some time now.  Early researchers in this literature focused on the cross-sectional relationship between prices and smoking.  By contrast, an influential study by Carpenter and Cook (2008) exploited within-state variation in taxes over time, which is now generally viewed as producing more reliable results.  Using YRBS data for the period 1991-2005, Carpenter and Cook (2008) found that a one-dollar increase in the per-pack cigarette tax was associated with a 3-6 percentage point reduction in smoking participation among high school students.

Our work was a follow-up to the Carpenter and Cook (hereafter C&C) study.  Since 2005, the last year of data available to C&C, 31 states increased their excise taxes by an average of 49 cents.  In addition, new anti-smoking strategies have been adopted, new tobacco products have come onto the market, and 4 additional waves of YRBS data have been collected (2007, 2009, 2011, and 2013).  Using these extra waves of data, we updated the C&C estimates of the relationship between cigarette taxes and youth smoking, and tested whether this relationship had changed over time.

We found that for the full period under study (1991-2013), a one-dollar increase in the cigarette tax was associated with a 1-3 percentage point reduction in youth smoking participation.  However, when we focused on the period 2007-2013, there was little evidence of a negative relationship between cigarette taxes and youth smoking (point estimates ranged from -.010 to 0.004 and were never statistically significant).  We speculated that anti-smoking efforts undertaken prior to 2005 may have discouraged all but the most price-insensitive youth from experimenting with tobacco, leaving a “hard-core” group of young people whose smoking initiation decisions are insensitive to tax-induced increases in the price of cigarettes.

This result illustrates the potential payoff to updating a well-known estimate in an established literature using more data.  In fact, we could not reject the hypothesis that the relationship between cigarette taxes and youth smoking was stable over the period 19991-2013 until we turned to state YRBS data, which contain information from over 800,000 respondents.  While it is valuable to confirm and extend the results obtained by previous researchers, meaningful policy variation and substantial sample sizes are often required in order to draw firm conclusions.

Finally, we explored the sensitivity of the C&C results (and our results) to including state-specific linear trends on the right-hand side of the regression model.  C&C did not include state-specific linear trends in any of their models, but one might legitimately worry that cigarette taxes could be capturing the effect of an omitted variable at the state level that trended smoothly over time such as anti-smoking sentiment. We found no evidence of a negative relationship between cigarette taxes and youth smoking using data for the period 1991-2013 with state-specific trends in the model.  Moreover, there was no evidence of a negative relationship between cigarette taxes and youth smoking in the period examined by C&C (1991-2005) with state-specific trends in the model.

We are not claiming that the specification with state-specific linear trends is correct and the C&C specification is incorrect.  Far from it. There are, of course, good reasons to include state-specific linear time trends on the right-hand side, but it is not at all clear what exactly they are controlling for.  What is clear, however, is that the wide range of estimates produced using different waves of YRBS data and across reasonable specification choices should give pause to anyone who is dead certain that the best way to discourage youth from smoking in the future is to impose additional cigarette taxes.

Reference

Carpenter, Christopher, and Philip J. Cook. 2008. “Cigarette Taxes and Youth Smoking: New Evidence from National, State, and Local Youth Risk Behavior Surveys. Journal of Health Economics, 27(2), 287-299.

1 See: The Tax Burden on Tobacco, Historical Compilation Volume 49, 2014